CNDB-12739: Fix row filter ignoring distinct index and query analyzers (

#1548)

Make RowFilter.Expression consider the two different index analyzers that an indexed column can have, 
one for write time (index_analyzer) and the other for read time (query_analyzer).

src/java/org/apache/cassandra/cql3/conditions/ColumnCondition.java

@@ -257,7 +257,7 @@ else if (otherValue == null)
                 // the condition value is not null, so only NEQ can return true
                 return operator == Operator.NEQ;
             }
-            return operator.isSatisfiedBy(type, otherValue, value, null); // We don't use any analyzers in LWT, see CNDB-11658
+            return operator.isSatisfiedBy(type, otherValue, value, null, null); // We don't use any analyzers in LWT, see CNDB-11658
         }
     }
 

src/java/org/apache/cassandra/cql3/conditions/ColumnConditions.java

@@ -93,7 +93,7 @@ public Set<ColumnMetadata> getAnalyzedColumns(IndexRegistry indexRegistry)
 
         for (ColumnCondition condition : this)
         {
-            if (indexRegistry.getAnalyzerFor(condition.column, condition.operator).isPresent())
+            if (indexRegistry.getIndexAnalyzerFor(condition.column, condition.operator).isPresent())
             {
                 analyzedColumns.add(condition.column);
             }

src/java/org/apache/cassandra/db/filter/RowFilter.java

@@ -261,7 +261,7 @@ public boolean partitionKeyRestrictionsAreSatisfiedBy(DecoratedKey key, Abstract
             ByteBuffer value = keyValidator instanceof CompositeType
                              ? ((CompositeType) keyValidator).split(key.getKey())[e.column.position()]
                              : key.getKey();
-            if (!e.operator().isSatisfiedBy(e.column.type, value, e.value, e.analyzer()))
+            if (!e.operator().isSatisfiedBy(e.column.type, value, e.value, e.indexAnalyzer(), e.queryAnalyzer()))
                 return false;
         }
         return true;
@@ -278,7 +278,7 @@ public boolean clusteringKeyRestrictionsAreSatisfiedBy(Clustering<?> clustering)
             if (!e.column.isClusteringColumn())
                 continue;
 
-            if (!e.operator().isSatisfiedBy(e.column.type, clustering.bufferAt(e.column.position()), e.value, e.analyzer()))
+            if (!e.operator().isSatisfiedBy(e.column.type, clustering.bufferAt(e.column.position()), e.value, e.indexAnalyzer(), e.queryAnalyzer()))
                 return false;
         }
         return true;
@@ -465,7 +465,7 @@ else if (builder.current.children.size() == 1 && builder.current.expressions.isE
         public SimpleExpression add(ColumnMetadata def, Operator op, ByteBuffer value)
         {
             assert op != Operator.ANN : "ANN expressions should be added with the addANNExpression method";
-            SimpleExpression expression = new SimpleExpression(def, op, value, analyzer(def, op), null);
+            SimpleExpression expression = new SimpleExpression(def, op, value, indexAnalyzer(def, op), queryAnalyzer(def, op), null);
             add(expression);
             return expression;
         }
@@ -479,18 +479,24 @@ public SimpleExpression add(ColumnMetadata def, Operator op, ByteBuffer value)
          */
         public void addANNExpression(ColumnMetadata def, ByteBuffer value, ANNOptions annOptions)
         {
-            add(new SimpleExpression(def, Operator.ANN, value, null, annOptions));
+            add(new SimpleExpression(def, Operator.ANN, value, null, null, annOptions));
         }
 
         public void addMapComparison(ColumnMetadata def, ByteBuffer key, Operator op, ByteBuffer value)
         {
-            add(new MapComparisonExpression(def, key, op, value, analyzer(def, op)));
+            add(new MapComparisonExpression(def, key, op, value, indexAnalyzer(def, op), queryAnalyzer(def, op)));
         }
 
         @Nullable
-        private Index.Analyzer analyzer(ColumnMetadata def, Operator op)
+        private Index.Analyzer indexAnalyzer(ColumnMetadata def, Operator op)
         {
-            return indexRegistry == null ? null : indexRegistry.getAnalyzerFor(def, op).orElse(null);
+            return indexRegistry == null ? null : indexRegistry.getIndexAnalyzerFor(def, op).orElse(null);
+        }
+
+        @Nullable
+        private Index.Analyzer queryAnalyzer(ColumnMetadata def, Operator op)
+        {
+            return indexRegistry == null ? null : indexRegistry.getQueryAnalyzerFor(def, op).orElse(null);
         }
 
         public void addGeoDistanceExpression(ColumnMetadata def, ByteBuffer point, Operator op, ByteBuffer distance)
@@ -870,7 +876,13 @@ public Operator operator()
         }
 
         @Nullable
-        public Index.Analyzer analyzer()
+        public Index.Analyzer indexAnalyzer()
+        {
+            return null;
+        }
+
+        @Nullable
+        public Index.Analyzer queryAnalyzer()
         {
             return null;
         }
@@ -1046,7 +1058,9 @@ public Expression deserialize(DataInputPlus in, int version, TableMetadata metad
                 ByteBuffer name = ByteBufferUtil.readWithShortLength(in);
                 Operator operator = Operator.readFrom(in);
                 ColumnMetadata column = metadata.getColumn(name);
-                Index.Analyzer analyzer = IndexRegistry.obtain(metadata).getAnalyzerFor(column, operator).orElse(null);
+                IndexRegistry indexRegistry = IndexRegistry.obtain(metadata);
+                Index.Analyzer indexAnalyzer = indexRegistry.getIndexAnalyzerFor(column, operator).orElse(null);
+                Index.Analyzer queryAnalyzer = indexRegistry.getQueryAnalyzerFor(column, operator).orElse(null);
 
                 // Compact storage tables, when used with thrift, used to allow falling through this withouot throwing an
                 // exception. However, since thrift was removed in 4.0, this behaviour was not restored in CASSANDRA-16217
@@ -1058,11 +1072,11 @@ public Expression deserialize(DataInputPlus in, int version, TableMetadata metad
                     case SIMPLE:
                         ByteBuffer value = ByteBufferUtil.readWithShortLength(in);
                         ANNOptions annOptions = operator == Operator.ANN ? ANNOptions.serializer.deserialize(in, version) : null;
-                        return new SimpleExpression(column, operator, value, analyzer, annOptions);
+                        return new SimpleExpression(column, operator, value, indexAnalyzer, queryAnalyzer, annOptions);
                     case MAP_COMPARISON:
                         ByteBuffer key = ByteBufferUtil.readWithShortLength(in);
                         ByteBuffer val = ByteBufferUtil.readWithShortLength(in);
-                        return new MapComparisonExpression(column, key, operator, val, analyzer);
+                        return new MapComparisonExpression(column, key, operator, val, indexAnalyzer, queryAnalyzer);
                     case VECTOR_RADIUS:
                         Operator boundaryOperator = Operator.readFrom(in);
                         ByteBuffer distance = ByteBufferUtil.readWithShortLength(in);
@@ -1119,26 +1133,40 @@ public long serializedSize(Expression expression, int version)
     public abstract static class AnalyzableExpression extends Expression
     {
         @Nullable
-        protected final Index.Analyzer analyzer;
+        protected final Index.Analyzer indexAnalyzer;
+
+        @Nullable
+        protected final Index.Analyzer queryAnalyzer;
 
-        public AnalyzableExpression(ColumnMetadata column, Operator operator, ByteBuffer value, @Nullable Index.Analyzer analyzer)
+        public AnalyzableExpression(ColumnMetadata column,
+                                    Operator operator,
+                                    ByteBuffer value,
+                                    @Nullable Index.Analyzer indexAnalyzer,
+                                    @Nullable Index.Analyzer queryAnalyzer)
         {
             super(column, operator, value);
-            this.analyzer = analyzer;
+            this.indexAnalyzer = indexAnalyzer;
+            this.queryAnalyzer = queryAnalyzer;
+        }
+
+        @Nullable
+        public final Index.Analyzer indexAnalyzer()
+        {
+            return indexAnalyzer;
         }
 
         @Nullable
-        public final Index.Analyzer analyzer()
+        public final Index.Analyzer queryAnalyzer()
         {
-            return analyzer;
+            return queryAnalyzer;
         }
 
         @Override
         public int numFilteredValues()
         {
-            return analyzer == null
+            return indexAnalyzer == null
                    ? super.numFilteredValues()
-                   : analyzer().analyze(value).size();
+                   : indexAnalyzer().analyze(value).size();
         }
     }
 
@@ -1153,10 +1181,11 @@ public static class SimpleExpression extends AnalyzableExpression
         public SimpleExpression(ColumnMetadata column,
                                 Operator operator,
                                 ByteBuffer value,
-                                @Nullable Index.Analyzer analyzer,
+                                @Nullable Index.Analyzer indexAnalyzer,
+                                @Nullable Index.Analyzer queryAnalyzer,
                                 @Nullable ANNOptions annOptions)
         {
-            super(column, operator, value, analyzer);
+            super(column, operator, value, indexAnalyzer, queryAnalyzer);
             this.annOptions = annOptions;
         }
 
@@ -1194,13 +1223,13 @@ public boolean isSatisfiedBy(TableMetadata metadata, DecoratedKey partitionKey,
                                 return false;
 
                             ByteBuffer counterValue = LongType.instance.decompose(CounterContext.instance().total(foundValue, ByteBufferAccessor.instance));
-                            return operator.isSatisfiedBy(LongType.instance, counterValue, value, analyzer);
+                            return operator.isSatisfiedBy(LongType.instance, counterValue, value, indexAnalyzer, queryAnalyzer);
                         }
                         else
                         {
                             // Note that CQL expression are always of the form 'x < 4', i.e. the tested value is on the left.
                             ByteBuffer foundValue = getValue(metadata, partitionKey, row);
-                            return foundValue != null && operator.isSatisfiedBy(column.type, foundValue, value, analyzer);
+                            return foundValue != null && operator.isSatisfiedBy(column.type, foundValue, value, indexAnalyzer, queryAnalyzer);
                         }
                     }
                 case NEQ:
@@ -1214,7 +1243,7 @@ public boolean isSatisfiedBy(TableMetadata metadata, DecoratedKey partitionKey,
                         assert !column.isComplex() : "Only CONTAINS and CONTAINS_KEY are supported for 'complex' types";
                         ByteBuffer foundValue = getValue(metadata, partitionKey, row);
                         // Note that CQL expression are always of the form 'x < 4', i.e. the tested value is on the left.
-                        return foundValue != null && operator.isSatisfiedBy(column.type, foundValue, value, analyzer);
+                        return foundValue != null && operator.isSatisfiedBy(column.type, foundValue, value, indexAnalyzer, queryAnalyzer);
                     }
                 case CONTAINS:
                     return contains(metadata, partitionKey, row);
@@ -1345,9 +1374,10 @@ public MapComparisonExpression(ColumnMetadata column,
                                        ByteBuffer key,
                                        Operator operator,
                                        ByteBuffer value,
-                                       @Nullable Index.Analyzer analyzer)
+                                       @Nullable Index.Analyzer indexAnalyzer,
+                                       @Nullable Index.Analyzer queryAnalyzer)
         {
-            super(column, operator, value, analyzer);
+            super(column, operator, value, indexAnalyzer, queryAnalyzer);
             assert column.type instanceof MapType && (operator == Operator.EQ || operator == Operator.NEQ || operator.isSlice());
             this.key = key;
         }

src/java/org/apache/cassandra/index/Index.java

@@ -431,12 +431,23 @@ default RowFilter.CustomExpression customExpressionFor(TableMetadata metadata, B
     }
 
     /**
-     * Returns the {@link Analyzer} for this index, if any. If the index doesn't transform the column values, this
-     * method will return an empty optional.
+     * Returns the write-time {@link Analyzer} for this index, if any. If the index doesn't transform the column values,
+     * this method will return an empty optional.
      *
-     * @return the transforming column value analyzer for the index, if any
+     * @return the write-time transforming column value analyzer for the index, if any
      */
-    default Optional<Analyzer> getAnalyzer()
+    default Optional<Analyzer> getIndexAnalyzer()
+    {
+        return Optional.empty();
+    }
+
+    /**
+     * Returns the query-time {@link Analyzer} for this index, if any. If the index doesn't transform the column values,
+     * this method will return an empty optional.
+     *
+     * @return the query-time transforming column value analyzer for the index, if any
+     */
+    default Optional<Analyzer> getQueryAnalyzer()
     {
         return Optional.empty();
     }

src/java/org/apache/cassandra/index/IndexRegistry.java

@@ -25,6 +25,7 @@
 import java.util.Optional;
 import java.util.Set;
 import java.util.concurrent.Callable;
+import java.util.function.Function;
 import java.util.function.Predicate;
 import java.util.function.Supplier;
 
@@ -314,13 +315,25 @@ default void registerIndex(Index index)
     Index getIndex(IndexMetadata indexMetadata);
     Collection<Index> listIndexes();
 
-    default Optional<Index.Analyzer> getAnalyzerFor(ColumnMetadata column, Operator operator)
+    default Optional<Index.Analyzer> getIndexAnalyzerFor(ColumnMetadata column, Operator operator)
+    {
+        return getAnalyzerFor(column, operator, Index::getIndexAnalyzer);
+    }
+
+    default Optional<Index.Analyzer> getQueryAnalyzerFor(ColumnMetadata column, Operator operator)
+    {
+        return getAnalyzerFor(column, operator, Index::getQueryAnalyzer);
+    }
+
+    default Optional<Index.Analyzer> getAnalyzerFor(ColumnMetadata column,
+                                                    Operator operator,
+                                                    Function<Index, Optional<Index.Analyzer>> analyzerGetter)
     {
         for (Index index : listIndexes())
         {
             if (index.supportsExpression(column, operator))
             {
-                Optional<Index.Analyzer> analyzer = index.getAnalyzer();
+                Optional<Index.Analyzer> analyzer = analyzerGetter.apply(index);
                 if (analyzer.isPresent())
                     return analyzer;
             }

src/java/org/apache/cassandra/index/sai/StorageAttachedIndex.java

@@ -660,26 +660,36 @@ public AbstractType<?> customExpressionValueType()
     }
 
     @Override
-    public Optional<Analyzer> getAnalyzer()
+    public Optional<Analyzer> getIndexAnalyzer()
     {
-        if (!indexContext.isAnalyzed())
-            return Optional.empty();
+        return indexContext.isAnalyzed()
+               ? Optional.of(value -> analyze(indexContext.getAnalyzerFactory(), value))
+               : Optional.empty();
+    }
 
-        return Optional.of(value -> {
-            List<ByteBuffer> tokens = new ArrayList<>();
-            AbstractAnalyzer analyzer = indexContext.getQueryAnalyzerFactory().create();
-            try
-            {
-                analyzer.reset(value);
-                while (analyzer.hasNext())
-                    tokens.add(analyzer.next());
-            }
-            finally
-            {
-                analyzer.end();
-            }
-            return tokens;
-        });
+    @Override
+    public Optional<Analyzer> getQueryAnalyzer()
+    {
+        return indexContext.isAnalyzed()
+               ? Optional.of(value -> analyze(indexContext.getQueryAnalyzerFactory(), value))
+               : Optional.empty();
+    }
+
+    private static List<ByteBuffer> analyze(AbstractAnalyzer.AnalyzerFactory factory, ByteBuffer value)
+    {
+        List<ByteBuffer> tokens = new ArrayList<>();
+        AbstractAnalyzer analyzer = factory.create();
+        try
+        {
+            analyzer.reset(value.duplicate());
+            while (analyzer.hasNext())
+                tokens.add(analyzer.next());
+        }
+        finally
+        {
+            analyzer.end();
+        }
+        return tokens;
     }
 
     @Override

src/java/org/apache/cassandra/index/sai/plan/Operation.java

@@ -311,7 +311,8 @@ static Node buildExpression(QueryController controller, RowFilter.Expression exp
                                                                                                         ByteBufferAccessor.instance,
                                                                                                         offset,
                                                                                                         ProtocolVersion.V3),
-                                                                               expression.analyzer(),
+                                                                               expression.indexAnalyzer(),
+                                                                               expression.queryAnalyzer(),
                                                                                expression.annOptions())));
                     offset += TypeSizes.INT_SIZE + ByteBufferAccessor.instance.getInt(expression.getIndexValue(), offset);
                 }

test/distributed/org/apache/cassandra/distributed/test/TestBaseImpl.java

@@ -32,6 +32,7 @@
 import org.junit.Assert;
 import org.junit.BeforeClass;
 
+import org.apache.cassandra.cql3.CQLTester;
 import org.apache.cassandra.db.marshal.*;
 import org.apache.cassandra.distributed.Cluster;
 import org.apache.cassandra.distributed.api.Feature;
@@ -180,4 +181,38 @@ void assertCannotStartDueToConfigurationException(Cluster cluster)
             Assert.assertEquals(ConfigurationException.class.getName(), tr.getClass().getName());
         }
     }
+
+    /**
+     * Runs the given function before and after a flush of sstables.  This is useful for checking that behavior is
+     * the same whether data is in memtables or sstables.
+     *
+     * @param cluster the tested cluster
+     * @param keyspace the keyspace to flush
+     * @param runnable the test to run
+     */
+    public static void beforeAndAfterFlush(Cluster cluster, String keyspace, CQLTester.CheckedFunction runnable) throws Throwable
+    {
+        try
+        {
+            runnable.apply();
+        }
+        catch (Throwable t)
+        {
+            throw new AssertionError("Test failed before flush:\n" + t, t);
+        }
+
+        for (int i = 1; i <= cluster.size(); i++)
+        {
+            cluster.get(i).flush(keyspace);
+
+            try
+            {
+                runnable.apply();
+            }
+            catch (Throwable t)
+            {
+                throw new AssertionError("Test failed after flushing node " + i + ":\n" + t, t);
+            }
+        }
+    }
 }